Group 3:¶

Predicting CTA Station Usage Using Ridership, Demographic, and Socioeconomic Data¶

Lucy Chavez, DePaul University, School of Computing, lchave28@depaul.edu¶

Denvir Gama, DePaul University, School of Computing, dlnu1@depaul.edu¶

Zach Hollis, DePaul University, School of Computing, zhollis@depaul.edu¶

Danielle Martin, DePaul University, School of Computing, dmart164@depaul.edu¶

In [38]:
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import colors
import seaborn as sns
from geopandas import GeoDataFrame
import geopandas as gpd
from mpl_toolkits.axes_grid1 import make_axes_locatable
from shapely.geometry import Point

sns.set_style('darkgrid')
In [3]:
sales_gdf = pd.read_csv('data/sales_gdf.csv')
census_gdf = gpd.read_file("https://data.cityofchicago.org/resource/bt9m-d2mf.geojson?$limit=9999999")
census_gdf["geo_12"] = census_gdf["geoid10"].map(lambda x: str(x)[:12])
neighborhood_gdf = gpd.read_file('../data/Boundaries - Neighborhoods.geojson')
In [5]:
final_df = pd.read_csv('data/full_data.csv')
final_df["geo_12"] = final_df["GEO_ID"].map(lambda x: str(x)[-12:])
final_df.info()
final_df.head(3)

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 218374 entries, 0 to 218373
Data columns (total 33 columns):
 #   Column                    Non-Null Count   Dtype  
---  ------                    --------------   -----  
 0   GEO_ID                    218353 non-null  object 
 1   Year                      218374 non-null  int64  
 2   pin                       218374 non-null  int64  
 3   sale_price                218374 non-null  float64
 4   age                       218374 non-null  int64  
 5   addr                      218371 non-null  object 
 6   hd_sf                     218374 non-null  float64
 7   n_units                   218374 non-null  float64
 8   bldg_sf                   132602 non-null  float64
 9   geometry                  218374 non-null  object 
 10  pri_neigh                 218374 non-null  object 
 11  sec_neigh                 218374 non-null  object 
 12  side                      218374 non-null  object 
 13  actual_sf                 218374 non-null  float64
 14  Total Population          186813 non-null  float64
 15  Median Age                186811 non-null  float64
 16  Median HH Income          183052 non-null  float64
 17  Total Housing Units       186813 non-null  float64
 18  Median Number of Rooms    186805 non-null  float64
 19  Median Year Built         173949 non-null  float64
 20  Median Gross Rent         171853 non-null  float64
 21  Mean HH Size              186777 non-null  float64
 22  Percent White             186813 non-null  float64
 23  Percent Black             186813 non-null  float64
 24  Percent HH with Children  186753 non-null  float64
 25  Percent Housing Vacant    186813 non-null  float64
 26  Number of Public Schools  186813 non-null  float64
 27  crime_count               186807 non-null  float64
 28  crimes_per_capita         186807 non-null  float64
 29  station_id                218345 non-null  float64
 30  station_name              218345 non-null  object 
 31  distance_miles            218345 non-null  float64
 32  geo_12                    218374 non-null  object 
dtypes: float64(22), int64(3), object(8)
memory usage: 55.0+ MB
Out[5]:
GEO_ID Year pin sale_price age addr hd_sf n_units bldg_sf geometry ... Percent Black Percent HH with Children Percent Housing Vacant Number of Public Schools crime_count crimes_per_capita station_id station_name distance_miles geo_12
0 1500000US170318392002 2015 17273100931118 53000.0 44 2921 S MICHIGAN AVE 365398.0 679.0 NaN POINT (-87.62257654590501 41.8402291575061) ... 0.699771 0.265432 0.111835 1.0 209.0 0.159664 30382.0 Cermak-McCormick Place 0.213983 170318392002
1 1500000US170313514001 2018 17343220150000 410000.0 130 3819 S WABASH AVE 3850.0 1.0 5050.0 POINT (-87.62436774377279 41.8249549567756) ... 0.917211 0.433090 0.062945 1.0 296.0 0.161220 30214.0 35th-Bronzeville-IIT 0.183507 170313514001
2 1500000US170318420001 2019 17341030641004 250000.0 11 223 E 31ST ST 2738.0 3.0 NaN POINT (-87.62080085952471 41.8381539098695) ... NaN NaN NaN NaN NaN NaN 30214.0 35th-Bronzeville-IIT 0.303191 170318420001

3 rows × 33 columns

In [6]:
final_df[final_df['pri_neigh'] == 'Loop'][['pri_neigh', 'sale_price', 'addr', 'actual_sf']].sort_values(by='actual_sf')
Out[6]:
pri_neigh sale_price addr actual_sf
178735 Loop 24000.0 235 W VAN BUREN ST 17.332037
180119 Loop 26000.0 235 W VAN BUREN ST 17.332037
180368 Loop 235000.0 235 W VAN BUREN ST 17.332037
179494 Loop 295000.0 235 W VAN BUREN ST 17.332037
180036 Loop 318500.0 235 W VAN BUREN ST 17.332037
... ... ... ... ...
178855 Loop 1200000.0 1142 S MICHIGAN AVE 1068.750000
177823 Loop 1278500.0 1142 S MICHIGAN AVE 1068.750000
180439 Loop 215000.0 533 S PLYMOUTH CT 1921.000000
180830 Loop 1925000.0 192 N PARK DR 18187.222222
180611 Loop 1875000.0 225 N COLUMBUS DR 18187.222222

3657 rows × 4 columns

In [7]:
final_df['pricer_per_sqft'] = final_df['sale_price'] / final_df['actual_sf']
#final_df['pricer_per_unit'] = final_df['sale_price'] / final_df['n_units']
final_df_avgsale = pd.DataFrame(final_df.groupby('pri_neigh')['pricer_per_sqft'].mean().sort_values(ascending=False)).reset_index()
final_df_avgsale = final_df_avgsale.merge(sales_gdf[['pri_neigh', 'side']].drop_duplicates(), on='pri_neigh', how='left')
final_df_avgsale[final_df_avgsale['side'] == 'Central']
Out[7]:
pri_neigh pricer_per_sqft side
0 Streeterville inf Central
1 Rush & Division 3952.285508 Central
2 Loop 3817.829633 Central
3 River North 3790.253389 Central
4 Gold Coast 2999.647596 Central
5 Printers Row 1788.759813 Central
8 Near South Side 1314.880008 Central

Data Visualization¶

In [8]:
final_df['pricer_per_unit'] = final_df['sale_price'] / (final_df['hd_sf'] / final_df['n_units'])

final_df_avgsale = pd.DataFrame(final_df.groupby('pri_neigh')['pricer_per_unit'].median().sort_values(ascending=False)).reset_index()
final_df_avgsale = final_df_avgsale.merge(sales_gdf[['pri_neigh', 'side']].drop_duplicates(), on='pri_neigh', how='left')

fig, ax = plt.subplots(figsize=(14,16))

sns.barplot(final_df_avgsale.pricer_per_unit, final_df_avgsale.pri_neigh, hue=final_df_avgsale.side, dodge=False)
ax.set_title('Median Sale Prices per Square Feet by Neighborhood (2013-2019)', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Median Sale Price', fontsize=15)
ax.set_ylabel('Neighborhood', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/saleprice_per_sqft_by_nbh.png')
plt.show()
In [9]:
final_df_avgsale = pd.DataFrame(final_df.groupby('pri_neigh')['sale_price'].median().sort_values(ascending=False)).reset_index()
final_df_avgsale = final_df_avgsale.merge(sales_gdf[['pri_neigh', 'side']].drop_duplicates(), on='pri_neigh', how='left')

fig, ax = plt.subplots(figsize=(14,16))

sns.barplot(final_df_avgsale.sale_price, final_df_avgsale.pri_neigh, hue=final_df_avgsale.side, dodge=False)
ax.set_title('Median Sale Prices by Neighborhood (2013-2019)', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Median Sale Price', fontsize=15)
ax.set_ylabel('Neighborhood', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/median_saleprice_by_nbh.png')
plt.show()
In [10]:
sales_compare = final_df.groupby(['pri_neigh', 'Year'])['sale_price'].median().unstack()
sales_compare['2013-2015'] = sales_compare[[2013, 2014, 2015]].mean(axis=1)
sales_compare['2016-2019'] = sales_compare[[2016, 2017, 2018, 2019]].mean(axis=1)
sales_compare = sales_compare[['2013-2015', '2016-2019']]
sales_compare = sales_compare.sort_values(by='2016-2019', ascending=True)


my_range=range(1,len(sales_compare.index)+1)
 
# The vertical plot is made using the hline function
# I load the seaborn library only to benefit the nice looking feature
fig, ax = plt.subplots(figsize=(14,16))

plt.hlines(y=my_range, xmin=sales_compare['2013-2015'], xmax=sales_compare['2016-2019'], color='grey', alpha=0.6)
plt.scatter(sales_compare['2013-2015'], my_range, color='skyblue', alpha=1, label='2013-2015')
plt.scatter(sales_compare['2016-2019'], my_range, color='green', alpha=0.6 , label='2016-2019')
plt.legend()
 
# Add title and axis names
ax.set_title('Comparison of Median Sales Prices in 2013-2015 and 2016-2019', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Median Sale Price', fontsize=15)
ax.set_ylabel('Neighborhood', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(my_range, sales_compare.index)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/median_saleprice_avg.png')
plt.show()
In [11]:
sales_compare_2 = final_df[(final_df['Year'] == 2013) | (final_df['Year'] == 2019)]
sales_compare_2 = sales_compare_2.groupby(['pri_neigh', 'Year'])['sale_price'].median().unstack()
sales_compare_2 = sales_compare_2.sort_values(by=2019, ascending=True)
sales_compare_2['diff'] = sales_compare_2[2019] - sales_compare_2[2013]


my_range=range(1,len(sales_compare_2.index)+1)
 
# The vertical plot is made using the hline function
# I load the seaborn library only to benefit the nice looking feature
fig, ax = plt.subplots(figsize=(14,16))

plt.hlines(y=my_range, xmin=sales_compare_2[2013], xmax=sales_compare_2[2019], color='grey', alpha=0.6)
plt.scatter(sales_compare_2[2013], my_range, color='skyblue', alpha=1, label='2013')
plt.scatter(sales_compare_2[2019], my_range, color='green', alpha=0.6 , label='2019')
plt.legend()
 
# Add title and axis names
ax.set_title('Comparison of Median Sales Prices in 2013 and 2019', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Median Sale Price', fontsize=15)
ax.set_ylabel('Neighborhood', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(my_range, sales_compare_2.index)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/median_saleprice.png')
plt.show()
In [12]:
median_income_by_year = final_df.groupby(['side', 'Year'])['Median HH Income'].median().unstack()
median_price_by_year = final_df.groupby(['side', 'Year'])['sale_price'].median().unstack()
median_inc_to_price = pd.DataFrame()
for year in range(2013, 2019):
    median_inc_to_price[year] = median_price_by_year[year] / median_income_by_year[year]
median_inc_to_price = median_inc_to_price.reset_index()
print(median_inc_to_price.head())

median_inc_to_price = median_inc_to_price.melt(id_vars=["side"], var_name="year", value_name="price_to_income_ratio")

sns.set_style('darkgrid')
fig, ax = plt.subplots(figsize=(15,8))

sns.lineplot(x="year", y="price_to_income_ratio", hue='side', data=median_inc_to_price, lw=3)
ax.set_title('Median Home Price to Income Ratio', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Year', fontsize=15)
ax.set_ylabel('Price to Income Ratio', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/side_price_to_income_ratio.png')
plt.show()

      side      2013      2014      2015      2016      2017      2018
0  Central  3.466284  3.434414  3.441419  3.319768  2.864886  2.858917
1    North  3.889184  4.290038  4.528302  4.528302  4.440913  4.227642
2    South  1.751501  2.087119  2.743069  2.777127  3.061474  3.190351
3     West  3.604115  4.224709  4.939682  4.823778  4.996813  4.964020
In [13]:
median_income_by_year = final_df.groupby(['pri_neigh', 'Year'])['Median HH Income'].median().unstack()
median_price_by_year = final_df.groupby(['pri_neigh', 'Year'])['sale_price'].median().unstack()
median_inc_to_price = pd.DataFrame()
for year in range(2013, 2019):
    median_inc_to_price[year] = median_price_by_year[year] / median_income_by_year[year]
median_inc_to_price = median_inc_to_price.reset_index()#.sort_values(by=2018, ascending=False)
median_inc_to_price['biggest_diff'] = (median_inc_to_price[2018] - median_inc_to_price[2013]) / median_inc_to_price[2013]
median_inc_to_price = median_inc_to_price.sort_values(by='biggest_diff', ascending=False)
median_inc_to_price = median_inc_to_price[['pri_neigh', 2013, 2014, 2015, 2016, 2017, 2018]]
print(median_inc_to_price.head())

median_inc_to_price = median_inc_to_price[:6].melt(id_vars=["pri_neigh"], var_name="year", value_name="price_to_income_ratio")

sns.set_style('darkgrid')
fig, ax = plt.subplots(figsize=(15,8))
#sns.palplot(sns.color_palette("Blues"))

sns.barplot(x="pri_neigh", y="price_to_income_ratio", hue="year", data=median_inc_to_price, palette="Blues")
#sns.barplot(x="year", y="price_to_income_ratio", hue="pri_neigh", data=median_inc_to_price)
#sns.lineplot(x="year", y="price_to_income_ratio", hue='pri_neigh', data=median_inc_to_price, lw=3)
ax.set_title('Highest Increase in Median Sale Price to Income Ratio', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Year', fontsize=15)
ax.set_ylabel('Price to Income Ratio', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='upper left', fontsize=15)

fig.tight_layout()
plt.savefig('figures/neighborhood_price_to_income_ratio.png')
plt.show()

        pri_neigh      2013      2014      2015      2016      2017      2018
63        Oakland  2.066203  2.300667  6.392447  4.417966  8.068127  9.681972
28    Fuller Park  0.834356  1.534884  1.017821  1.505665  1.614661  3.456749
21        Douglas  1.356222  2.812280  3.159225  3.600510  2.819977  4.189535
31  Garfield Park  1.484146  1.671034  2.579560  2.885574  3.448540  4.560394
57       New City  0.997755  1.347826  1.629265  2.002470  2.174163  2.903740
In [ ]:
In [14]:
sale_pop = final_df.groupby(['pri_neigh', 'side']).agg({'sale_price':'median', 'Percent White':'median', 'pin':'count'})
sale_pop = sale_pop.reset_index().rename(columns={"pin": "count"})
sale_pop = sale_pop.sort_values(by='count', ascending=False)
print(sale_pop.head())

#sns.set(style="white")
fig, ax = plt.subplots(figsize=(15,9))

sns.scatterplot(x="Percent White", y="sale_price", hue="side", size="count",
                sizes=(10, 2000), alpha=.5, palette="muted", data=sale_pop)
ax.set_title('Median Residential Sales Price by Community Demographic (2013-2019)', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Percent White', fontsize=15)
ax.set_ylabel('Median Sales Price', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='upper left', fontsize=15, ncol=2, frameon=False)

fig.tight_layout()
plt.savefig('figures/demographic_scatter.png')
plt.show()

       pri_neigh   side  sale_price  Percent White  count
44     Lake View  North    371000.0       0.877754  12894
45  Lincoln Park  North    500000.0       0.879563   6530
81        Uptown  North    235000.0       0.713781   5340
65  Portage Park  North    265255.5       0.700915   5146
25     Edgewater  North    183000.0       0.708833   5025
In [15]:
# sns.set(style="white", rc={"axes.facecolor": (0, 0, 0, 0)})

# # Initialize the FacetGrid object
# pal = sns.cubehelix_palette(10, rot=-.25, light=.7)
# g = sns.FacetGrid(df, row="Year", hue="Year", aspect=5.5, height=2.2, palette=pal)

# # Draw the densities in a few steps
# g.map(sns.kdeplot, "sale_price", clip_on=False, shade=True, alpha=1, lw=10.5, bw=2)
# g.map(sns.kdeplot, "sale_price", clip_on=False, color="w", lw=2, bw=.2)
# g.map(plt.axhline, y=0, lw=2, clip_on=False)


# # Define and use a simple function to label the plot in axes coordinates
# def label(x, color, label):
#     ax = plt.gca()
#     ax.text(0.9, .1, label, fontweight="bold", color=color, fontsize=30,
#             ha="left", va="center", transform=ax.transAxes)


# g.map(label, "sale_price")

# # Set the subplots to overlap
# g.fig.subplots_adjust(hspace=-.8)

# # Remove axes details that don't play well with overlap
# g.set_titles("")
# g.set(yticks=[])
# g.despine(bottom=True, left=True)

# plt.savefig('figures/overlap_density.png')
# plt.show()
In [16]:
sales_gdf_count = pd.DataFrame(sales_gdf.groupby('pri_neigh')['pin'].nunique().sort_values(ascending=False)).reset_index()
sales_gdf_count = sales_gdf_count.merge(sales_gdf[['pri_neigh', 'side']].drop_duplicates(), on='pri_neigh', how='left')

sns.set_style('darkgrid')
fig, ax = plt.subplots(figsize=(14,16))

sns.barplot(sales_gdf_count.pin, sales_gdf_count.pri_neigh, hue=sales_gdf_count.side, dodge=False)
ax.set_title('Residential Sales by Neighborhood (2013-2019)', fontsize=25,
             fontweight="bold", loc='center', y=1.01)
ax.set_xlabel('Number of Residential Property Sales', fontsize=15)
ax.set_ylabel('Neighborhood', fontsize=15)
plt.xticks(fontsize=13)
plt.yticks(fontsize=13)
plt.legend(loc='lower right', fontsize=15)

fig.tight_layout()
plt.savefig('figures/num_sales.png')
plt.show()
In [17]:
sales_compare = final_df.groupby(['geo_12', 'Year'])['sale_price'].median().unstack()
sales_compare['2013-2015'] = sales_compare[[2013, 2014, 2015]].mean(axis=1)
sales_compare['2016-2019'] = sales_compare[[2016, 2017, 2018, 2019]].mean(axis=1)
sales_compare['sales_price_diff'] = (sales_compare['2016-2019'] - sales_compare['2013-2015'])  / sales_compare['2013-2015']
sales_compare = sales_compare.reset_index()
sales_compare = sales_compare.sort_values(by='sales_price_diff')[:-6]
sales_compare
Out[17]:
Year geo_12 2013 2014 2015 2016 2017 2018 2019 2013-2015 2016-2019 sales_price_diff
359 170310810005 NaN 6.586000e+06 NaN 645000.0 NaN 545000.000000 NaN 6.586000e+06 595000.000000 -0.909657
353 170310804002 NaN NaN 1150000.0 NaN 565000.00000 25000.000000 25000.0 1.150000e+06 205000.000000 -0.821739
1080 170313510003 442000.0 NaN 465000.0 115000.0 85000.00000 128900.000000 145000.0 4.535000e+05 118475.000000 -0.738754
1626 170316708002 87500.0 9.200000e+04 115000.0 19000.0 30100.00000 45000.000000 37500.0 9.816667e+04 32900.000000 -0.664856
292 170310701021 188000.0 1.000500e+06 50000.0 137194.0 175000.00000 192000.000000 181700.0 4.128333e+05 171473.500000 -0.584642
... ... ... ... ... ... ... ... ... ... ... ...
968 170312606001 16000.0 2.055000e+04 33000.0 135000.0 54900.00000 88000.000000 105500.0 2.318333e+04 95850.000000 3.134436
1185 170314313011 24250.0 2.000000e+04 23300.0 45400.0 113500.00000 65700.000000 152900.0 2.251667e+04 94375.000000 3.191340
1239 170314603023 34121.5 2.650000e+04 20450.0 133000.0 135000.00000 147314.850717 42500.0 2.702383e+04 114453.712679 3.235288
1003 170312912001 36750.0 3.200000e+04 21000.0 87500.0 160500.00000 101750.000000 220000.0 2.991667e+04 142437.500000 3.761142
1030 170313012002 NaN 2.652124e+04 NaN 112000.0 83360.24735 159000.000000 155000.0 2.652124e+04 127340.061837 3.801436

2144 rows × 11 columns

In [18]:
blockgroups = census_gdf.dissolve(by='geo_12').reset_index()
blockgroups.crs = "EPSG:4326"
merged_gdf = sales_compare[['geo_12', 'sales_price_diff']]\
    .merge(blockgroups[['geo_12', 'geometry']], on='geo_12', how="left")
merged_gdf = gpd.GeoDataFrame(merged_gdf.drop_duplicates()).sort_values(by='sales_price_diff')
print(merged_gdf.iloc[:,:2].head())

fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

norm = colors.TwoSlopeNorm (vmin=merged_gdf.sales_price_diff.min(),
                            vcenter=-0.2,
                            vmax=merged_gdf.sales_price_diff.max())
ax = merged_gdf.plot(column='sales_price_diff', legend=True, cmap='coolwarm',
                     ax=ax, cax=cax, norm=norm)
ax.set_title('Percent Change of Median Sales Prices in 2013-2015 and 2016-2019',
             fontsize=25, fontweight="bold", loc='center', y=2.12)

fig.tight_layout()
plt.savefig('figures/censusblock_saleprice_change_heatmap.png')
plt.show()

         geo_12  sales_price_diff
0  170310810005         -0.909657
1  170310804002         -0.821739
2  170313510003         -0.738754
3  170316708002         -0.664856
4  170310701021         -0.584642
In [19]:
sales_compare = final_df.groupby(['pri_neigh', 'Year'])['sale_price'].median().unstack()
sales_compare['2013-2015'] = sales_compare[[2013, 2014, 2015]].mean(axis=1)
sales_compare['2016-2019'] = sales_compare[[2016, 2017, 2018, 2019]].mean(axis=1)
sales_compare['sales_price_diff'] = (sales_compare['2016-2019'] - sales_compare['2013-2015'])  / sales_compare['2013-2015']

sales_compare = sales_compare.reset_index()
sales_compare = sales_compare[['pri_neigh', 'sales_price_diff']]\
    .merge(neighborhood_gdf[['pri_neigh', 'geometry']], on='pri_neigh', how="left")
sales_compare = gpd.GeoDataFrame(sales_compare.drop_duplicates()).sort_values(by='sales_price_diff')
print(sales_compare.iloc[:,:2].head())

fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

norm = colors.TwoSlopeNorm (vmin=-0.5,
                            vcenter=0,
                            vmax=sales_compare.sales_price_diff.max())
ax = sales_compare.plot(column='sales_price_diff', legend=True, cmap='coolwarm', ax=ax, cax=cax, norm=norm)
ax.set_title('Percent Change of Median Sales Prices in 2013-2015 and 2016-2019', fontsize=25,
             fontweight="bold", loc='center', y=2.12)

fig.tight_layout()
plt.savefig('figures/saleprice_change_heatmap.png')
plt.show()

        pri_neigh  sales_price_diff
64       Old Town         -0.066361
50           Loop         -0.033756
78  Streeterville         -0.022544
44      Lake View          0.001512
33     Gold Coast          0.014450
In [20]:
median_income_by_year = final_df.groupby(['pri_neigh', 'Year'])['Median HH Income'].median().unstack()
median_price_by_year = final_df.groupby(['pri_neigh', 'Year'])['sale_price'].median().unstack()
median_inc_to_price = pd.DataFrame()
for year in range(2013, 2019):
    median_inc_to_price[year] = median_price_by_year[year] / median_income_by_year[year]
median_inc_to_price = median_inc_to_price.reset_index()#.sort_values(by=2018, ascending=False)
median_inc_to_price['biggest_diff'] = (median_inc_to_price[2018] - median_inc_to_price[2013]) / median_inc_to_price[2013]
median_inc_to_price = median_inc_to_price.sort_values(by='biggest_diff', ascending=False)
median_inc_to_price = median_inc_to_price[['pri_neigh', 2013, 2014, 2015, 2016, 2017, 2018]]

median_inc_to_price['2013-2015'] = median_inc_to_price[[2013, 2014, 2015]].mean(axis=1)
median_inc_to_price['2016-2018'] = median_inc_to_price[[2016, 2017, 2018]].mean(axis=1)
median_inc_to_price['price_income_diff'] = (median_inc_to_price['2016-2018'] - median_inc_to_price['2013-2015'])\
    / median_inc_to_price['2013-2015']
median_inc_to_price = median_inc_to_price[['pri_neigh', 'price_income_diff']]\
    .merge(neighborhood_gdf[['pri_neigh', 'geometry']], on='pri_neigh', how="left")
median_inc_to_price = gpd.GeoDataFrame(median_inc_to_price.drop_duplicates()).sort_values(by='price_income_diff')
print(median_inc_to_price.iloc[:, :2].head())

fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

norm = colors.TwoSlopeNorm (vmin=-0.9,
                            vcenter=0,
                            vmax=median_inc_to_price.price_income_diff.max())
ax = median_inc_to_price.plot(column='price_income_diff', legend=True, cmap='coolwarm', ax=ax, cax=cax, norm=norm)
ax.set_title('Percent Change of Median Sale Price to \n Income Ratio in 2013-2015 and 2016-2019', fontsize=25,
             fontweight="bold", loc='center', y=2.12)

fig.tight_layout()
plt.savefig('figures/saleprice_income_ratio_change_heatmap.png')
plt.show()

          pri_neigh  price_income_diff
92     Printers Row          -0.300500
90  Near South Side          -0.181503
82     North Center          -0.164953
83      River North          -0.148524
84        Greektown          -0.142458

PREDICTIONS¶

In [36]:
df_20_21 = pd.read_csv('data/2020_2021_predictions.csv')
df_2020 = df_20_21.loc[df_20_21['Year'] == 2020]



df_2021 = df_20_21.loc[df_20_21['Year'] == 2021]
df_2021
Out[36]:
Unnamed: 0 GEO_ID Year predicted_price_p_house
1 1 1500000US170310101001 2021 278222.678931
3 3 1500000US170310101002 2021 211273.287766
5 5 1500000US170310101003 2021 267228.114437
7 7 1500000US170310102011 2021 282613.433289
9 9 1500000US170310102012 2021 221563.790953
... ... ... ... ...
4400 4400 1500000US170318439001 2021 192989.742540
4402 4402 1500000US170318439002 2021 245046.684180
4404 4404 1500000US170318439003 2021 133450.854427
4406 4406 1500000US170318439004 2021 136565.792758
4408 4408 1500000US170318439005 2021 137152.774347

2212 rows × 4 columns

In [37]:
blockgroups = census_gdf.dissolve(by='geo_12').reset_index()
blockgroups.crs = "EPSG:4326"

df_2019 = pd.read_csv('data/2019_predictions.csv')
df_2019["geo_12"] = df_2019["GEO_ID"].map(lambda x: str(x)[-12:])
df_2019 = df_2019[['Year', 'actual_price_p_house', 'predicted_price_p_house', 'geo_12']]\
    .merge(blockgroups[['geo_12', 'geometry']], on='geo_12', how="left")
df_2019 = gpd.GeoDataFrame(df_2019.drop_duplicates())
df_2019_block = gpd.sjoin(df_2019, neighborhood_gdf, how='inner', op='intersects')
df_2019_neigh = gpd.sjoin(neighborhood_gdf, df_2019, how='inner', op='intersects')

df_20_21 = pd.read_csv('data/2020_2021_predictions.csv')

df_20_21 = pd.read_csv('data/2020_2021_predictions.csv')
df_2020 = df_20_21.loc[df_20_21['Year'] == 2020]
df_2020["geo_12"] = df_2020["GEO_ID"].map(lambda x: str(x)[-12:])
df_2020 = df_2020[['Year', 'predicted_price_p_house', 'geo_12']]\
    .merge(blockgroups[['geo_12', 'geometry']], on='geo_12', how="left")
df_2020 = gpd.GeoDataFrame(df_2020.drop_duplicates())
df_2020_block = gpd.sjoin(df_2020, neighborhood_gdf, how='inner', op='intersects')
df_2020_neigh = gpd.sjoin(neighborhood_gdf, df_2020, how='inner', op='intersects')

df_2021 = df_20_21.loc[df_20_21['Year'] == 2021]
df_2021["geo_12"] = df_2021["GEO_ID"].map(lambda x: str(x)[-12:])
df_2021 = df_2021[['Year', 'predicted_price_p_house', 'geo_12']]\
    .merge(blockgroups[['geo_12', 'geometry']], on='geo_12', how="left")
df_2021 = gpd.GeoDataFrame(df_2021.drop_duplicates())
df_2021_block = gpd.sjoin(df_2021, neighborhood_gdf, how='inner', op='intersects')
df_2021_neigh = gpd.sjoin(neighborhood_gdf, df_2021, how='inner', op='intersects')

C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(None != epsg:4326)
  "(%s != %s)" % (left_df.crs, right_df.crs)
C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(epsg:4326 != None)
  "(%s != %s)" % (left_df.crs, right_df.crs)
C:\Users\ymeri\anaconda3\lib\site-packages\ipykernel_launcher.py:17: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(None != epsg:4326)
  "(%s != %s)" % (left_df.crs, right_df.crs)
C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(epsg:4326 != None)
  "(%s != %s)" % (left_df.crs, right_df.crs)
C:\Users\ymeri\anaconda3\lib\site-packages\ipykernel_launcher.py:26: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(None != epsg:4326)
  "(%s != %s)" % (left_df.crs, right_df.crs)
C:\Users\ymeri\anaconda3\lib\site-packages\geopandas\tools\sjoin.py:61: UserWarning: CRS of frames being joined does not match!(epsg:4326 != None)
  "(%s != %s)" % (left_df.crs, right_df.crs)
In [182]:
# df_dict = {df_2019_block: ['actual_price_p_house', 'predicted_price_p_house'],
#            df_2019_neigh: ['actual_price_p_house', 'predicted_price_p_house'],
#            df_2020_block: ['predicted_price_p_house'],
#            df_2020_neigh: ['predicted_price_p_house']}

# for df_year in [(df, col) for df, cols in df_dict.items() for col in cols]:
#     for col in ['actual_price_p_house', 'predicted_price_p_house']:
#         fig, ax = plt.subplots(1, 1, figsize=(13,20))
#         divider = make_axes_locatable(ax)
#         cax = divider.append_axes("right", size="5%", pad=0.1)

#         ax = df_year.plot(column=col, legend=True, cmap='Reds', ax=ax, cax=cax)
#         ax.set_title('{}, {}'.format(df.Year[0], col.split('_')[0]), fontsize=25,
#              fontweight="bold", loc='center', y=2.12)
        
#         fig.tight_layout()
#         plt.savefig('figures/{}_{}_census_block.png'.format(df.Year[0], col.split('_')[0]))
#         plt.show()
In [41]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_actual_census_block'
ax = df_2019_block.plot(column='actual_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2019/{}.png'.format(table_name))
plt.show()
In [42]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_actual_neighborhood'
ax = df_2019_neigh.plot(column='actual_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2019/{}.png'.format(table_name))
plt.show()
In [44]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_predicted_census_block'
ax = df_2019_block.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2019/{}.png'.format(table_name))
plt.show()
In [48]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_predicted_neighborhood'
ax = df_2019_neigh.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2019/{}.png'.format(table_name))
plt.show()
In [56]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2020_predicted_census_block'
ax = df_2020_block.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2020_2021/{}.png'.format(table_name))
plt.show()
In [57]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2020_predicted_neighborhood'
ax = df_2020_neigh.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2020_2021/{}.png'.format(table_name))
plt.show()
In [71]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

# norm = colors.TwoSlopeNorm (vmin=-0.1,
#                             vcenter=0,
#                             vmax=1.5)                              
table_name = '2021_predicted_census_block'
ax = df_2021_block.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2020_2021/{}.png'.format(table_name))
plt.show()
In [59]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2021_predicted_neighborhood'
ax = df_2021_neigh.plot(column='predicted_price_p_house', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2020_2021/{}.png'.format(table_name))
plt.show()
In [60]:
price_diff = df_2019_block[['geo_12', 'actual_price_p_house']]\
    .merge(df_2021_block[['geo_12', 'predicted_price_p_house', 'geometry']], on='geo_12', how='inner')
price_diff = price_diff.rename(columns={'actual_price_p_house': '2019_actual_p',
                                                    'predicted_price_p_house': '2021_pred_p'})
price_diff['price_diff_perc'] = (price_diff['2021_pred_p'] - price_diff['2019_actual_p']) / price_diff['2019_actual_p']
price_diff = gpd.GeoDataFrame(price_diff.drop_duplicates())
price_diff.loc[price_diff['price_diff_perc'] > 6, 'price_diff_perc'] = 6
price_diff.sort_values(by='price_diff_perc', ascending=False)
Out[60]:
geo_12 2019_actual_p 2021_pred_p geometry price_diff_perc
2773 170310804002 2.500000e+04 260958.243589 POLYGON ((-87.64112 41.90375, -87.64121 41.903... 6.000000
2811 170310812011 1.300000e+05 966501.705910 POLYGON ((-87.62693 41.89916, -87.62719 41.899... 6.000000
7604 170318400002 6.577610e+04 465555.462049 POLYGON ((-87.63862 41.82355, -87.63886 41.823... 6.000000
8483 170314314004 1.852500e+04 131014.677122 POLYGON ((-87.55664 41.76238, -87.55645 41.762... 6.000000
2812 170310812011 1.300000e+05 866783.689662 POLYGON ((-87.62693 41.89916, -87.62719 41.899... 5.667567
... ... ... ... ... ...
3745 170310811002 7.445000e+05 332647.688410 POLYGON ((-87.62822 41.89752, -87.62822 41.897... -0.553193
6916 170313108002 7.750000e+05 341962.192567 POLYGON ((-87.66612 41.85689, -87.66612 41.856... -0.558758
10062 170316805002 1.555800e+05 64239.828844 POLYGON ((-87.65347 41.78690, -87.65371 41.786... -0.587095
3770 170310812012 9.083400e+05 360255.000911 POLYGON ((-87.62820 41.89668, -87.62820 41.896... -0.603392
2560 170310802021 1.103500e+06 358347.315565 POLYGON ((-87.63285 41.90388, -87.63300 41.903... -0.675263

2297 rows × 5 columns

In [63]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_2021_pred_price_diff_census_block'
ax = price_diff.plot(column='price_diff_perc', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/{}.png'.format(table_name))
plt.show()
In [61]:
actual_p_neigh = pd.DataFrame(df_2019_neigh.groupby('pri_neigh')['actual_price_p_house'].mean()).reset_index()
predicted_p_neigh = pd.DataFrame(df_2021_neigh.groupby('pri_neigh')['predicted_price_p_house'].mean()).reset_index()
neigh_p_diff = actual_p_neigh.merge(predicted_p_neigh)
neigh_p_diff['p_diff'] = (neigh_p_diff['predicted_price_p_house'] - neigh_p_diff['actual_price_p_house'])\
    / neigh_p_diff['actual_price_p_house']
neigh_p_diff = neigh_p_diff[['pri_neigh', 'p_diff']].merge(df_2021_neigh[['pri_neigh', 'geometry']])
neigh_p_diff = gpd.GeoDataFrame(neigh_p_diff.drop_duplicates())
neigh_p_diff.sort_values(by='p_diff', ascending=False)
Out[61]:
pri_neigh p_diff geometry
1120 Fuller Park 0.538875 MULTIPOLYGON (((-87.62880 41.80189, -87.62879 ...
2843 Riverdale 0.513056 MULTIPOLYGON (((-87.60157 41.68621, -87.60158 ...
500 Boystown 0.310065 MULTIPOLYGON (((-87.64878 41.93999, -87.64927 ...
1279 Gold Coast 0.255307 MULTIPOLYGON (((-87.62646 41.91147, -87.62640 ...
2783 Pullman 0.251163 MULTIPOLYGON (((-87.60157 41.68621, -87.60168 ...
... ... ... ...
2639 Oakland -0.032194 MULTIPOLYGON (((-87.59215 41.81693, -87.59231 ...
1858 Lincoln Park -0.033014 MULTIPOLYGON (((-87.62640 41.91137, -87.62646 ...
790 Chinatown -0.057517 MULTIPOLYGON (((-87.63022 41.85770, -87.63022 ...
3026 Sheffield & DePaul -0.111765 MULTIPOLYGON (((-87.65833 41.92166, -87.65835 ...
1728 Kenwood -0.127178 MULTIPOLYGON (((-87.59215 41.81693, -87.59215 ...

98 rows × 3 columns

In [62]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2019_2021_pred_price_diff_neighborhood'
ax = neigh_p_diff.plot(column='p_diff', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/{}.png'.format(table_name))
plt.show()

predicted price to income ratio¶

In [139]:
price_to_income_block.isnull().sum().sort_values(ascending = False)
Out[139]:
price_to_income            99
2018                       99
geometry                    0
predicted_price_p_house     0
geo_12                      0
dtype: int64
In [149]:
price_by_block = pd.DataFrame(df_2021_block.groupby('geo_12')['predicted_price_p_house'].mean().reset_index())
price_by_block = price_by_block.merge(df_2021_block[['geo_12', 'geometry']].drop_duplicates())
income_by_block = final_df.groupby(['geo_12', 'Year'])['Median HH Income'].median().unstack().reset_index()\
    [['geo_12', 2018]]
price_to_income_block = income_by_block.merge(price_by_block)
price_to_income_block['price_to_income'] = price_to_income_block['predicted_price_p_house'] / price_to_income_block[2018]
price_to_income_block = gpd.GeoDataFrame(price_to_income_block.drop_duplicates())
price_to_income_block.sort_values(by='price_to_income', ascending=False)
Out[149]:
geo_12 2018 predicted_price_p_house geometry price_to_income
2069 170318411004 14288.0 468416.784542 POLYGON ((-87.63004 41.84718, -87.63011 41.847... 32.783929
1077 170313511001 10077.0 302994.715768 POLYGON ((-87.61443 41.82751, -87.61465 41.827... 30.067948
20 170310105031 17227.0 466117.693154 POLYGON ((-87.66061 41.99863, -87.66059 41.998... 27.057392
228 170310605001 34130.0 805670.320152 POLYGON ((-87.66110 41.95102, -87.66145 41.951... 23.605928
1074 170313406001 12713.0 298524.377995 POLYGON ((-87.63324 41.82362, -87.63334 41.823... 23.481820
... ... ... ... ... ...
1978 170318356001 NaN 128787.630721 POLYGON ((-87.63371 41.80180, -87.63412 41.801... NaN
2018 170318382002 NaN 488402.657520 POLYGON ((-87.66641 41.86686, -87.66689 41.866... NaN
2025 170318387004 NaN 143112.187702 POLYGON ((-87.71643 41.85896, -87.71667 41.858... NaN
2091 170318421003 NaN 226954.742976 POLYGON ((-87.72979 41.89576, -87.72979 41.895... NaN
2130 170318439002 NaN 245046.684180 POLYGON ((-87.58632 41.77426, -87.58632 41.774... NaN

2134 rows × 5 columns

In [163]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2021 Price-to-income Ratio by Census Block'
ax = price_to_income_block.plot(column='price_to_income', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2021_price_to_income/{}.png'.format(table_name))
plt.show()
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [ ]:
In [161]:
price_by_neigh = pd.DataFrame(df_2021_neigh.groupby('pri_neigh')['predicted_price_p_house'].mean().reset_index())
price_by_neigh = price_by_neigh.merge(df_2021_neigh[['pri_neigh', 'geometry']].drop_duplicates())
income_by_neigh = final_df.groupby(['pri_neigh', 'side', 'Year'])['Median HH Income'].median().unstack().reset_index()\
    [['pri_neigh', 'side', 2018]]
price_to_income_neigh = income_by_neigh.merge(price_by_neigh)
price_to_income_neigh['price_to_income'] = price_to_income_neigh['predicted_price_p_house'] / price_to_income_neigh[2018]
price_to_income_neigh = gpd.GeoDataFrame(price_to_income_neigh.drop_duplicates())
price_to_income_neigh.sort_values(by='price_to_income', ascending=False).head(10)
Out[161]:
pri_neigh side 2018 predicted_price_p_house geometry price_to_income
3 Armour Square South 34303.0 387554.005203 MULTIPOLYGON (((-87.62920 41.84713, -87.62919 ... 11.297962
19 Chinatown South 42386.0 410604.371566 MULTIPOLYGON (((-87.63022 41.85770, -87.63022 ... 9.687264
63 Oakland South 30469.0 290000.733906 MULTIPOLYGON (((-87.59215 41.81693, -87.59231 ... 9.517895
80 United Center West 45771.0 423232.999226 MULTIPOLYGON (((-87.66707 41.88885, -87.66707 ... 9.246750
51 Lower West Side West 43006.0 387448.393777 MULTIPOLYGON (((-87.63516 41.85772, -87.63532 ... 9.009171
28 Fuller Park South 24011.0 206206.686494 MULTIPOLYGON (((-87.62880 41.80189, -87.62879 ... 8.588009
39 Humboldt Park West 41500.0 346761.092846 MULTIPOLYGON (((-87.74060 41.88782, -87.74060 ... 8.355689
12 Bridgeport South 46397.0 375172.910875 MULTIPOLYGON (((-87.64173 41.84728, -87.64173 ... 8.086146
8 Avondale North 59745.0 471878.329997 MULTIPOLYGON (((-87.68799 41.93610, -87.68798 ... 7.898206
91 Woodlawn South 29970.0 227180.637146 MULTIPOLYGON (((-87.58630 41.77334, -87.58664 ... 7.580268
In [164]:
fig, ax = plt.subplots(1, 1, figsize=(13,20))
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)

table_name = '2021 Price-to-income Ratio by Neighborhood'
ax = price_to_income_neigh.plot(column='price_to_income', legend=True, cmap='Reds', ax=ax, cax=cax)
ax.set_title(table_name, fontsize=25, fontweight="bold", loc='center', y=2.12)
fig.tight_layout()
plt.savefig('figures/2021_price_to_income/{}.png'.format(table_name))
plt.show()